Notes on routing, redirects and sample quality
Practical guides for researchers, panel operators and developers who move respondents between systems.
-
Pre-survey fraud screening: what it catches, what it misses and what it costs
Screening respondents before the survey opens is the cheapest place to remove fraud. It is not free, it is not complete, and it creates false positives you have to plan for.
-
AI agents now pass attention checks. Plan your fieldwork accordingly
A research team built an AI respondent that passed 99.8% of attention checks. That does not mean online sample is finished. It means the checks most studies rely on are aimed at the wrong layer.
-
Sample exchange or single panel? A candid comparison from a routing company
We run a router, so you would expect us to say exchanges win. For some studies a single good panel is the better choice, and we say which ones.
-
Incidence rate, LOI and CPI: how online sample is priced, and why quotes go wrong
Three numbers set the price of almost every online sample quote. Industry benchmark data shows the first one is routinely sold higher than it turns out to be. What each number means and how to estimate it without optimism.
-
Server-to-server postbacks vs browser redirects: which one should you trust?
Survey outcomes travel through the respondent's browser or directly between servers. Each channel breaks in its own way. Use redirects to move people and server calls to move money.
-
Ghost completes: the reconciliation problem nobody budgets for
A ghost complete is a complete your sample platform recorded and your survey data does not contain. They are invisible until someone matches IDs, and by then the invoice has usually been approved.
-
Survey redirect status codes: mapping outcomes across sample platforms
Every sample platform encodes survey outcomes its own way, and a surprising amount of reconciliation pain is just translation errors between them. Map outcomes, not numbers.
-
What is a survey endlink? Complete, terminate, over quota and quality redirects
Endlinks decide what gets paid for. Most reconciliation disputes we would expect to see trace back to four of them being set up carelessly, usually in the last hour before launch.
-
The GDQ removal codes, and why free-text reversal reasons have to go
Reconciliation has run on spreadsheets of IDs and reasons like "bad data" for as long as online sample has existed. The Global Data Quality code frame gives buyers and suppliers 18 shared reasons instead. This is how we would use them.
-
A survey data quality checklist for 2026: before, during and after fieldwork
The controls worth having in an online survey this year, sorted by where they run. Short on theory, specific about settings, and honest about which checks have stopped doing their job.
-
Monitor before you enforce: turning on fraud screening without wrecking fieldwork
The fastest way to lose a good supplier is to switch on blocking and watch their traffic disappear without an explanation. Run screening in monitor mode first, join the flags to your own cleaning results, and decide from evidence.
-
Why was I screened out of a survey? Terminates, full surveys and rejected responses explained
Being turned away a few questions into a survey is frustrating, and the explanations you usually get are vague. This is what actually happened, why it happens so often, and what you can do about it.